VoiceChat

Create a room: Enter your name on the lobby page and click Create New Room. You'll get a 6-character room code.

Join a room: Enter your name and the 6-character room code, then click Join Room. Or use a shared room link directly.

Voice controls: Use the Mute button to silence your mic. Use Deafen to mute all incoming audio. Enable Push-to-Talk in settings to transmit only while holding the button/spacebar.

Share a link: Click the room code to copy a shareable link to your clipboard.

Open AI Chat: Click the AI button in room controls (or the floating 🤖 button on mobile).

Select a provider: In the AI Chat → Settings tab, choose a provider:

• Local / Ollama: Runs models on the server. No API key needed. Models: llama3.2:1b, gemma:2b, phi3:mini.
• OpenRouter: Access to many models. Requires an API key (or use your own).
• Claude / OpenAI: Requires your own API key.

API Keys: Enter your key in the AI Settings tab. Keys are saved in your browser's localStorage — never sent to our server.

Slash commands: Type these in the AI chat input:

• /ai-reset — Clear conversation history
• /ai-help — Show available commands
• /ai-model <name> — Switch model (e.g. /ai-model llama3.2:1b)
• /ai-provider <name> — Switch provider (e.g. /ai-provider local)

Start a conversation: Go to the AI Chat → Voice tab and tap Start Conversation.

Status pills:

• 🎧 Listening — Recording your voice
• 🗣️ Speaking — AI is responding via text-to-speech
• 🧠 Thinking — AI is processing your request
• 🎤 Mic Off — Your microphone is muted

Auto-interrupt: Enable in AI settings to stop the AI mid-sentence by speaking.

Speech settings: Adjust silence threshold, speech rate, pitch, and volume in AI Settings.

Shows real-time connection information. Open it with the 📊 button in room controls.

Indicators:

• WebSocket: Connection to the signaling server
• ICE Gathering: Discovering network paths to peers
• Connection: WebRTC peer connection state
• TURN Relay: Whether traffic is relayed through a TURN server
• Peers: Number of connected participants
• Microphone: Your mic status (active/muted/inactive)
• Audio I/O: Your selected input/output devices

Colors: ● Green = good, ● Yellow = in progress, ● Red = error, ● Gray = idle/unknown

Frontend: Vanilla JavaScript, WebRTC (RTCPeerConnection), WebSocket API, Web Speech API (STT/TTS)

Backend: Node.js + Express + ws (WebSocket)

NAT Traversal: Google STUN + coturn TURN server

AI: Multi-provider — Ollama (local), OpenRouter, Claude, OpenAI

Deployment: Docker Compose + Traefik reverse proxy + Let's Encrypt TLS

Topology: Mesh (each peer connects to every other peer directly)

Browser support: Chrome, Firefox, Edge, Safari (iOS requires user gesture for microphone)

🎤 Microphone not working:

• Check browser permissions — click the lock icon in the address bar
• Make sure no other app is using the microphone
• Try refreshing the page
• Check that the correct input device is selected in Settings

🔌 Can't connect to room:

• Check your internet connection
• Try enabling "Force TURN" in settings (helps with restrictive NATs)
• The room may have been empty and cleaned up — create a new one

🤖 AI not responding:

• Check your API key in AI Settings
• Verify the provider is reachable (try a different one)
• Check the browser console for errors (enable Debug Logging in settings)

📡 Connection lost:

• The app will auto-reconnect with exponential backoff
• If reconnection fails, leave and rejoin the room
• Check the Live Status Panel for specific error details